Hi Steve,

I think Arndt is right, mkgmap requires much more heap now when computing
the index, also without the new --x-split-name-index option.
This is his error message:
Exception in thread "main" java.lang.OutOfMemoryError: GC overhead limit exceeded
        at java.nio.CharBuffer.wrap(Unknown Source)
        at java.nio.CharBuffer.wrap(Unknown Source)
        at uk.me.parabola.imgfmt.app.srt.Sort.createSortKey(Sort.java:182)
        at uk.me.parabola.imgfmt.app.srt.Sort.createSortKey(Sort.java:271)
        at uk.me.parabola.imgfmt.app.mdr.Mdr7.preWriteImpl(Mdr7.java:168)
        at uk.me.parabola.imgfmt.app.mdr.MdrSection.preWrite(MdrSection.java:129)
        at uk.me.parabola.imgfmt.app.mdr.MDRFile.writeSections(MDRFile.java:308)
        at uk.me.parabola.imgfmt.app.mdr.MDRFile.write(MDRFile.java:247)
        at uk.me.parabola.mkgmap.combiners.MdrBuilder.onFinishForDevice(MdrBuilder.java:355)
        at uk.me.parabola.mkgmap.combiners.GmapsuppBuilder.onFinish(GmapsuppBuilder.java:163)
        at uk.me.parabola.mkgmap.main.Main.endOptions(Main.java:575)
        at uk.me.parabola.mkgmap.CommandArgsReader.readArgs(CommandArgsReader.java:128)
        at uk.me.parabola.mkgmap.main.Main.mainStart(Main.java:134)
        at uk.me.parabola.mkgmap.main.Main.main(Main.java:105)

I see two reasons:
1) The structure MDR7Record is a bit larger
2) A lot more bytes are stored in byte[]  arrays and SrtSortKey instances

For a map for Austria this requires 169MB instead of 128 MB.
I think one reason is that Mdr7.preWriteImpl() is always computing and
saving a partial name, no matter what option is used?

Gerd